Handwritten English, Marathi and Sanskrit Word Recognition using Artificial Neural Network and Feature Extraction
Character recognition plays an important role in many recent applications. In recent times among worldwide
researchers there is increasing trend to identify handwritten words of many languages and scripts. Various feature extraction
techniques are used such as Stroke method, Fourier descriptor, Gradient feature extraction, and chain code histogram. The
recognition ratio can be increased by a proper feature extraction technique. Curvelet transform supports curve as well as
edge discontinuities. Several group of curvelet coefficient are generated at different scale after the curvelet transform and
angles. It is found that Neural Network gives enhanced accuracy for recognition purpose.In this an effort is made to
recognize handwritten characters for English alphabets also with feature extraction using Multilayer Feed Forward
Neural Network. Each character data set contains 26 alphabets and primarily 2 symbols viz. full stop & comma. 50 diverse
character data sets are used for training the neural network. In the proposed system, each character is resized into 30x20
pixels, which will directly be used for training. That is each resized character has 600 pixels and these pixels are taken
as features for training the neural network.
Keywords - Curvelet Transform, Neural Network (MLP), Multilayer Feed forward Network.